Chaotic dot-superconductor analog of the Hanbury Brown Twiss effect

P. Samuelsson,M. Buttiker
DOI: https://doi.org/10.1103/PhysRevLett.89.046601
2002-03-08
Abstract:As an electrical analog of the optical Hanbury Brown Twiss effect, we study current cross-correlations in a chaotic quantum dot-superconductor junction. One superconducting and two normal reservoirs are connected via point contacts to a chaotic quantum dot. For a wide range of contact widths and transparencies, we find large positive current correlations. The positive correlations are generally enhanced by normal backscattering in the contacts. Moreover, for normal backscattering in the contacts, the positive correlations survive when suppressing the proximity effect in the dot with a weak magnetic field.
Mesoscale and Nanoscale Physics,Superconductivity
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the behavior of current cross - correlations in quantum - dot - superconducting junctions, especially the magnitude of these correlations in multimode junctions and whether the proximity effect is required to produce positive current cross - correlations. Specifically, the author studied a system consisting of a chaotic quantum dot, a superconductor and two normal - metal reservoirs connected by point contacts. ### Main questions: 1. **Magnitude of positive correlations**: Is the positive current cross - correlation a significant effect, and is its magnitude proportional to the number of transmission modes in the contacts? 2. **Role of the proximity effect**: Is the proximity effect a necessary condition for obtaining positive correlations? ### Research background: - **Hanbury Brown - Twiss effect**: This is an optical interference experiment that demonstrates the Bose - statistical properties of photons. Similarly, the behavior of electrons in mesoscopic multi - terminal conductors can be analyzed in a similar way. - **Andreev reflection**: When a normal conductor is connected to a superconductor, correlations are generated between electrons and holes due to Andreev reflection, and this phenomenon is called the proximity effect. - **Previous research**: Previous studies have shown that positive correlations can be predicted in single - mode junctions, but in multi - terminal diffusive junctions, the cross - correlations are usually negative without the proximity effect. ### Research methods: The author used Random Matrix Theory (RMT) to describe the scattering matrix of the chaotic quantum dot and calculated the current cross - correlations in different situations to answer the above questions. They considered the cases of ideal and non - ideal contacts and explored the influence of the proximity effect on the correlations. ### Main conclusions: 1. **Significant positive correlations**: The positive current cross - correlation is indeed a significant effect, and its magnitude is proportional to the number of transmission modes. 2. **Proximity effect not required**: Positive correlations can exist without the proximity effect, especially in the case of non - ideal contacts, where normal backscattering enhances the positive correlations. 3. **Influence of normal backscattering**: Normal backscattering can lead to the transition from negative to positive correlations in some cases, especially when strongly coupled to normal reservoirs. ### Mathematical expressions: The expression for current cross - correlations is: \[ P_{12} = \frac{V^4 e^3}{h} \sum_{i,j = 1,2} \text{tr} \left[ (S_{\alpha e}^{1i})^\dagger S_{\alpha h}^{1j} (S_{\beta h}^{2j})^\dagger S_{\beta e}^{2i} \right] \] where \( \sigma = +(-) \) for \( \alpha = \beta (\alpha \neq \beta) \). In the case of ideal contacts, the average value of the total current correlation is: \[ \langle P_{12} \rangle_U = \frac{V^4 e^3}{h} \left( \text{tr}[R(1 - R)] - \frac{\text{tr}(R) \text{tr}(1 - R)}{4N} \right) \] The final normalized correlation expression is: \[ \frac{\langle P_{12} \rangle}{P_0} = \frac{(1 + \gamma)^2}{2\gamma^2} \left( \frac{1}{1 + \gamma} - \frac{2\gamma}{f} - \frac{f^2 - 16\gamma^3}{f^{5/2}} \right) \] where \( f(\gamma)=1 + 6\gamma+\gamma^2 \) and \( \gamma = \frac{2N}{M} \). These results indicate that positive correlations not only exist significantly, but can also be enhanced under certain conditions, even without the proximity effect.